Plasticity of Chasmogamous and Cleistogamous Reproductive Allocation in Grasses Gregory P

Aliso: A Journal of Systematic and Evolutionary Botany

Volume 23 | Issue 1 Article 23

2007 Plasticity of Chasmogamous and Cleistogamous Reproductive Allocation in Grasses Gregory P. Cheplick College of Staten Island, City University of New York, Staten Island

Follow this and additional works at: http://scholarship.claremont.edu/aliso Part of the Botany Commons, and the Ecology and Evolutionary Biology Commons

Recommended Citation Cheplick, Gregory P. (2007) "Plasticity of Chasmogamous and Cleistogamous Reproductive Allocation in Grasses," Aliso: A Journal of Systematic and Evolutionary Botany: Vol. 23: Iss. 1, Article 23. Available at: http://scholarship.claremont.edu/aliso/vol23/iss1/23 Aliso 23, pp. 286–294 ᭧ 2007, Rancho Santa Ana Botanic Garden

PLASTICITY OF CHASMOGAMOUS AND CLEISTOGAMOUS REPRODUCTIVE ALLOCATION IN GRASSES

GREGORY P. CHEPLICK Department of Biology, College of Staten Island, City University of New York, Staten Island, New York 10314, USA ([email protected])

ABSTRACT Cleistogamy is more common in grasses than in any other angiosperm family. Both self-fertilized cleistogamous (CL) spikelets and open-pollinated chasmogamous (CH) spikelets are typically produced. Relative allocation to CL and CH varies among species and populations, and is influenced by ontogeny and environment. The balance between reproductive modes can be expressed as a CH/CL ratio. This ratio is very plastic and stressful conditions can result in values Ͻ1.0. In Amphicarpum purshii, an annual with subterranean CL spikelets, CH/CL declined as density increased because CH decreased more than CL as size was reduced by intraspecific competition. In the shade-tolerant annual Microstegium vimineum, CH/CL was lowest in large greenhouse-grown plants in an unlimited, sunny environment, but was highest in small plants from a shady forest interior; tiller vegetative mass often showed a negative allometric relation to CH and CL allocation. In the perennial Dichanthelium clandestinum, CH and CL allocation varied among populations, but there was no consistent effect of light on CH/CL. The phenology of reproduction strongly affects CH/CL. In Danthonia spicata, CH/CL was high early in the season as CH flowering commenced, but dropped quickly as axillary CL spikelets matured; A. purshii showed the opposite pattern because CL reproduction occurred first. The assumption that cleistogamy simply provides reproductive assurance should be reevaluated in light of new information on phenology and allometry. Changes in the balance between CH and CL caused by environmental factors may be indirect effects of size. Evolutionary models that do not explore the plasticity and allometry of CH and CL reproduction may not be useful in predicting the myriad patterns in nature. Key words: allometry, chasmogamy, cleistogamy, grasses, phenotypic plasticity, Poaceae, reproductive allocation.

INTRODUCTION of CL may be difficult to detect in the field and have prob- ably led to underestimates of the extent of self-fertilization Cleistogamy (CL) is the production of flowers that do not in grasses. Other types of CL are more readily observed by open and undergo self-fertilization ‘‘in the bud’’ (Lord casual inspection. ‘‘Sheath fertilization’’ involves the reten- 1981). Most species that have CL flowers can and do con- tion of spikelets on non-emergent, axillary inflorescences en- tinue to make open, potentially cross-pollinated chasmoga- closed within leaf sheaths and is the most common type of mous (CH) flowers. Because an individual usually can make CL in grasses (Campbell et al. 1983). Depending on the both floral forms and they may be present at the same time, species, axillary spikelets may be found all along the stem CL species are often considered to have a mixed breeding axis or be mostly concentrated at the basal nodes (Dykster- system or ‘‘multiple strategy’’ (Lloyd 1984; Schoen 1984; huis 1945; Dobrenz and Beetle 1966; Cheplick and Clay Redbo-Torstensson and Berg 1995; Masuda et al. 2001). Due 1989; Cheplick 1996). In a few grasses, CL spikelets and to the structural and developmental differences typically found between CH and CL flowers (and sometimes, between caryopses (seeds) mature at the ends of specialized culms the resulting fruits and seeds), CL species are also consid- beneath the soil surface (Campbell et al. 1983; Cheplick ered to show reproductive dimorphism (Lord 1981; Clay 1994). 1983a; Schoen and Lloyd 1984; Cheplick 1994; Plitmann The relative extent of CL can vary greatly both between 1995). and within phylogenetically related species (Campbell In Poaceae, CL is more common than in any other angio- 1982). Perhaps the best example is Clay’s (1983a) detailed sperm family and has been reported in over 320 species comparison of multiple populations of four species of Dan- (Campbell et al. 1983). Cleistogamy may be due to spikelets thonia DC. Percent CL was defined as the number of CL that do not open because of modified floral parts, precocious spikelets divided by the number of CH ϩ CL spikelets. maturation while retained within enclosing leaf sheaths, or Hence, % CL shows the fraction of the total flowers that lodicule failure (Campbell 1982; Heslop-Harrison and Hes- were CL at the time the population was sampled. For five lop-Harrison 1996). For example, Campbell (1982) de- populations of D. sericea Nutt., mean % CL (Ϯ SE) was scribed how, in some species of Andropogon L., spikelets only 2.7 Ϯ 0.3%, while for two populations of D. compressa within the raceme sheath can go through anthesis and self- Austin % CL was 50.2 Ϯ 1.0%. Populations were most var- fertilize before the inflorescence fully emerges from the en- iable for D. spicata (L.) P. Beauv., ranging from 5 to 43% closing sheath. Without detailed observation, the above types CL (x Ϯ SE ϭ 24.5 Ϯ 0.4%). Because higher % CL in D. VOLUME 23 Chasmogamy/Cleistogamy in Grasses 287 spicata was associated with disturbed and mown habitats; grasses (Clay 1982, 1983a; Bell and Quinn 1987; Cheplick Clay (1983a) suggested grazing pressure might generally fa- 1995). Environmental factors known to affect the extent of vor increased levels of CL in grasses, a view shared by oth- CL in grasses include light, soil nutrition and moisture, and ers (Dyksterhuis 1945; Campbell et al. 1983). competition and herbivory. Because CL flowers are energy Other research has indicated that the extent of CL in efficient for a plant to produce and provide reproductive as- grasses can vary among seed-derived sibships (Clay 1982; surance (Schemske 1978; Schoen 1984; Schoen and Lloyd Cheplick and Quinn 1988) and among clonal replicates in 1984), it has long been noted that whenever environmental perennials (Cheplick 1995). In Amphicarpum purshii Kunth, conditions are stressful there tends to be greater reproduction an annual, the ratio of seeds from CH spikelets to those from by CL compared to CH (Dyksterhuis 1945; Wilken 1982; CL spikelets had a significant narrow-sense heritability of Campbell et al. 1983). Hence, the ratio of CH to CL is ex- 0.56 (Cheplick and Quinn 1988). For the perennial Dan- pected to decline with increasing environmental stress. It is thonia spicata, Clay (1982) reported broad-sense heritabili- also recognized that both CH and CL depend on ontogeny, ties of 0.53 and 0.72 in the field and greenhouse, respec- varying with plant size and phenological stages in ways that tively, for the proportion of CL spikelets produced per tiller. may be predictable for some species (Clay 1982; Jasieniuk Although simple recessive inheritance of CL (one to three and Lechowicz 1987; Cheplick and Clay 1989; Cheplick genes) has been reported in some grain crops (Merwine et 1994; Diaz and Macnair 1998). As will be explored later in al. 1981; Chhabra and Sethi 1991; Kurauchi et al. 1993; this paper, the challenge for the future generation of models Ueno and Itoh 1997), like most life-history traits in wild of the evolution of mixed CH/CL breeding systems will be species, CL is likely to be under quantitative genetic control to account for the marked plasticity of the two reproductive (Mazer and LeBuhn 1999). Given the primacy of genetic modes in relation to environmental factors. In addition, they variation to the microevolution of populations, it is surpris- will need to explore the relations of plant size and pheno- ing that so little effort has gone into documenting quantita- logical stage to observed changes in the extent of CL shown tive genetic variation in the extent of CL vs. CH in different during plant development. species. Objectives Theory Given the predictions made by theoretical models for the Theoretical considerations of the evolution of CL have evolution of CH/CL and the complexities inherent in trying involved comparative cost-benefit analyses of CH vs. CL to understand life-history traits that vary greatly with envi- flower production (Schoen and Lloyd 1984) or variation in ronmental conditions, one objective of this paper will be to the relative fertility of the two floral types (Redbo-Torstens- summarize information on the relative allocation to CH and son and Berg 1995; Masuda et al. 2001). Because the pa- CL in a variety of grasses exposed to variable environments. ternal costs of producing prodigious amounts of pollen in The balance between reproductive modes will be expressed outcrossed spikelets may be high in grasses (e.g., Schoen as a CH/CL ratio. Because many of the changes in this bal- 1984), maternal reallocation of conserved resources to CL ance that are mediated by environmental factors may be the spikelets could enhance plant fitness by increasing the num- indirect effect of plant size, a second objective will be to ber or size of maturing seeds (Schoen and Lloyd 1984; present allometric relationships of CH and CL to vegetative Lloyd 1987). Indeed, seeds of CL spikelets are heavier than mass for several grass species for which data are available. those of CH spikelets in a number of grass species (Camp- A final objective is to describe how temporal changes in CH/ bell et al. 1983; Schoen 1984; Cheplick 1994). Assuming no CL are related to the phenology of the two reproductive inbreeding depression, this can translate into more successful modes. establishment of seedlings that germinate from the seeds of CL spikelets (Cheplick and Quinn 1983; Clay 1983b; MATERIALS AND METHODS Schoen 1984). Data Sources The production of CL flowers might compensate for the loss in fruit set due to the unsuccessful fertilization of many Information on CH and CL allocation expressed as a pro- CH flowers, providing ‘‘reproductive assurance’’ (Redbo- portion of plant (or tiller) dry mass in relation to environ- Torstensson and Berg 1995; Culley 2002). Where environmental factors was obtained from published sources. Ratios ments fluctuate between conditions that are favorable and of CH/CL were calculated from allocation or dry mass data. unfavorable for each reproductive mode, CL may be envis- Allometric relationships between CH or CL allocation and aged as a ‘‘bet-hedging strategy’’ (e.g., Berg and Redbo- vegetative mass were characterized by regression analysis

Torstensson 1998). The relative balance between CH and CL and results are depicted in conventional log10–log10 plots should depend on the ability of a plant to assess and respond (Niklas 1994). Vegetative mass is the sum of root and shoot to environmental variation (Schoen and Lloyd 1984). In the mass. model presented by Masuda et al. (2001), seasonal variation The following environmental factors and grasses were ex- in the fertility of CH flowers was used to explain why some amined: density for Amphicarpum purshii (Cheplick 1982; species switch between reproductive modes during the Cheplick and Quinn 1983), soil nutrients for A. purshii (Che- course of the growing season. plick 1989), and Triplasis purpurea (Walter) Chapm. (Che- There is no question that the relative production of CH plick 1996), light and soil moisture for six Dichanthelium and CL flowers, fruits, and seeds is markedly plastic in most clandestinum (L.) Gould populations (Bell and Quinn 1987), facultatively CL plants (e.g., Schemske 1978), including and light for Microstegium vimineum (Trin.) A. Camus (Che- 288 Cheplick ALISO plick unpubl. data; see below). In addition, phenological data Table 1. Dry mass (mg) of vegetative (root ϩ shoot) parts, and on CH and CL allocation were used to depict temporal CL and CH spikelets and seeds of Amphicarpum purshii reared in changes in CH/CL across a single growing season in Am- a greenhouse at variable densities. Pots were 11.4 cm in diameter. phibromus scabrivalvis (Trin.) Swallen (Cheplick and Clay Data adapted from Cheplick (1982). Values are x Ϯ SE; N ϭ 10. Coefficients of variation are in parentheses. 1989), Amphicarpum purshii (Cheplick 1982), and Danthon- ia spicata (Cheplick and Clay 1989). Density Vegetative CL CH

Data for Microstegium vimineum.—Japanese stiltgrass is an Control (1/pot) 722.9 Ϯ 61.4 92.2 Ϯ 11.0 56.3 Ϯ 4.8 Asian species that has become very invasive in the eastern (26.8) (37.8) (26.9) USA (Hunt and Zaremba 1992; Gibson et al. 2002). Al- Low (5/pot) 338.7 Ϯ 20.5 44.0 Ϯ 2.2 11.5 Ϯ 1.3 (19.1) (15.5) (34.6) though it is a summer annual C4 grass, it is shade tolerant (Horton and Neufeld 1998) and tends to grow in mesic de- Medium (15/pot) 216.5 Ϯ 4.6 31.4 Ϯ 1.4 2.4 Ϯ 0.3 (6.7) (14.2) (43.4) ciduous forests where it can form a dense, lawn-like mono- High (30/pot) 174.7 Ϯ 8.4 27.1 Ϯ 1.7 1.1 Ϯ 0.2 culture underneath the canopy. In temperate regions, it re- (4.8) (19.3) (62.0) cruits from seed in early spring and produces culms that can branch and root from the lowermost nodes. Terminal racemes that bear CH spikelets emerge in late summer. The RESULTS spikelets undergo anthesis and mature seed in early autumn. In addition, CL spikelets form on sheath-enclosed axillary Density racemes beginning at the uppermost nodes in late summer. Plasticity of CH and CL allocation in relation to density Mature seeds from CH and CL spikelets are present simul- was determined from an experiment with the amphicarpic taneously on individual plants in early autumn. annual Amphicarpum purshii (Cheplick and Quinn 1983), a A large stand within a forest in Millstone Township, cen- species with subterranean CL spikelets and aerial, terminal tral New Jersey, USA, was selected for detailed study. Com- CH spikelets (Cheplick 1994). In an intraspecific competi- plete tillers with mature seeds were collected every 20 m tion experiment, plants were reared in a greenhouse at four along a transect through the shady forest understory on 5 densities: 1, 5, 15, and 30 per 11.4 cm diameter pot. These Oct 2001 until seeds from 20 plants had been sampled. All will be referred to as control, low, medium, and high den- seeds were stratified at 4ЊC for four months on moistened sities, respectively, and correspond to 98.0, 489.9, 1469.6, filter paper in petri dishes. Seedlings obtained from these and 2939.2 plants/m2. These treatments nicely bracket the seeds were used to establish a greenhouse population of M. natural range in density from 152/m2 to 2808/m2 (calculated vimineum on 3 May 2002. There were 120 plants, 6 per from data in Cheplick 1982). The experiment lasted three family. Details of this experiment are reported elsewhere months and there were ten replications of each density (Che- (Cheplick 2005). For the present purpose, it should be rec- plick and Quinn 1983). At harvest, despite a decrease in the ognized that these plants experienced a high-resource envi- dry mass of both CH and CL reproductive components with ronment, growing in high light and supplied with mineral increasing density (Table 1), CL allocation was remarkably nutrients (fertilizer) on two occasions. As terminal racemes constant, varying only from 13 to 16% (Fig. 1). In contrast, with seeds of CH spikelets matured (4 Sep–15 Oct 2002), a CH allocation declined from 8% in the control to 0.6% at single, large senesced tiller was sampled. Tiller length, num- the highest density. Thus, CH/CL showed a precipitous de- ber of branches, and culm, leaf, terminal CH, and axillary crease with increasing density (Fig. 2). CL seed dry mass (after two days at 60ЊC) were obtained. Because A. purshii size was also reduced significantly Data were collected for 20 randomly selected individuals with density (Table 1; Cheplick and Quinn 1983), the allom- (one per maternal family). etry of CH/CL to vegetative dry mass (i.e., root ϩ shoot To compare the very large plants from the high-resource, mass) was determined across densities. The relationship of greenhouse environment to plants in the more stressful nat- CH/CL to vegetative mass (VM) was best described by a ural environment, two tiller populations were sampled at the polynomial regression (Fig. 3): log(CH/CL) ϭϪ26.0 ϩ 17.8 field site in autumn 2002, when tillers bore mature CH and log VM Ϫ 3.1 log(VM)2 (N ϭ 30, r2 ϭ 0.84, P Ͻ 0.001). axillary CL racemes. On 15 Oct, tillers were collected from In this species, CH/CL tends to increase up to a point, but the forest edge where plants experienced full sun (1100– then levels off to where the allocation to CH is about 60% 1800 ␮mol/m2/s) for at least 2–4 hr/day. On 24 Oct, 20 ad- that of allocation to CL in the largest plants (Fig. 3). It is ditional tillers were collected from within the shaded forest also noteworthy that CH allocation was more variable than interior where light levels during the previous growing sea- CL allocation at low, medium, and especially high density, son (barring sunflecks) were only 5–30 ␮mol/m2/s for the as assessed by coefficients of variation (Table 1). entire day (Cheplick unpubl. data). These tillers from the ‘‘edge’’ and ‘‘shaded’’ populations Soil Fertility were dried for two days at 60ЊC and data were recorded as Plasticity of the balance in CH and CL in relation to soil for the greenhouse population described above. It should be fertility was explored for Triplasis purpurea, an annual with noted that the genetic background for all three experimental axillary CL. In this greenhouse study, plants were exposed groups (greenhouse, edge, shaded) should be similar because to three treatments: low nutrients (water only), medium nu- all plants sampled had been part of the large original M. trients (1.5 g/liter 20-20-20 N-P-K fertilizer) and high nutri- vimineum stand at the field site. ents (3 g/liter 20-20-20 N-P-K fertilizer). Details of the VOLUME 23 Chasmogamy/Cleistogamy in Grasses 289

Fig. 3.—Allometry of CH/CL (log10–log10 plot) in the intraspeciﬁc competition experiment with Amphicarpum purshii. Symbols show plants from control (ⅷ), low (Ⅵ), medium (᭡), and high (᭢) density treatments (see Cheplick and Quinn 1983).

data, CH/CL ratios were calculated. There was no signiﬁcant difference in CH/CL in high nutrients (0.2468 Ϯ 0.0207) vs. low nutrients (0.2289 Ϯ 0.0106) (t ϭ 0.76, P Ͼ 0.5).

Light and Soil Moisture In a detailed study of the effects of soil moisture and light intensity on the plasticity of CH and CL allocation patterns, Fig. 1, 2.—Relation of CH and CL allocation, and CH/CL, to Bell and Quinn (1987) investigated six populations of the intraspecific density in Amphicarpum purshii. All bars are x Ϯ SE; N ϭ 10.—1. Percent allocation of vegetative mass to CH and CL.— caespitose perennial Dichanthelium clandestinum. Seeds 2. Ratio of CH to CL. were collected from six sites in central New Jersey that differed in soil moisture and light levels. Some plants were reared in a greenhouse along a soil moisture gradient, while growing conditions are available in Cheplick (1996). Al- others were subjected to one of three light treatments: high though plant size data were not recorded, the number of (68–86% full sun), medium (31–40% full sun), or low (11– seeds matured in CH and CL spikelets and their dry masses 14% full sun). Allocation to CL was always much greater were determined. The mean total number of seeds/tiller, % than allocation to CH in all treatments. Across the soil mois- CL seeds/tiller, and CH/CL mass ratio are shown in Table 2. ture gradient, allocation to CH remained relatively constant, Not surprisingly, higher nutrient levels enhanced seed output but was Ͻ1% in all populations. However, the proportion of (for log10 transformed no. seeds, F ϭ 8.49, P Ͻ 0.001); how- the reproductive allocation comprising CL flowers increased ever, the percent of seeds matured in CL spikelets was high as soil moisture decreased, in accordance with the expecta- (75–81%) regardless of nutrient conditions (for arcsine, tion of greater CL under stressful conditions. Nevertheless, square-root transformed proportion of CL seeds, F ϭ 1.34, it may be suspected that the changes in CH/CL across the P Ͼ 0.2; Table 2). With medium or high levels of nutrients, moisture gradient were simply indirect effects of plant size CH/CL showed no significant increase relative to low nutrients (for arcsine, square-root transformed CH/CL, F ϭ 0.53, Table 2. Mean (Ϯ SE) number of seeds per tiller, percent of P Ͼ 0.5). seeds from CL spikelets, and CH/CL mass ratio for Triplasis pur- In a study of the effects of nutrient availability in Amphi- purea reared in a greenhouse at three levels of soil nutrients (N ϭ carpum purshii, plants were subjected to high (3 g/liter 20- 13 per nutrient level). See Cheplick 1996. 20-20 N-P-K fertilizer) or low (water only) nutrients and harvested when senesced (Cheplick 1989). Although high Nutrients No. seeds % CL CH/CL nutrients resulted in significantly greater vegetative mass (x Ϯ SE: 384.1 Ϯ 9.7 mg in high vs. 287.3 Ϯ 15.6 mg in low Low 44.75 Ϯ 4.67 80.95 Ϯ 2.98 0.2678 Ϯ 0.0631 nutrients), the dry mass of CL and CH seeds was not cor- Medium 80.92 Ϯ 6.43 74.59 Ϯ 3.01 0.3439 Ϯ 0.0591 High 78.00 Ϯ 9.12 77.64 Ϯ 2.69 0.3012 Ϯ 0.0422 related with vegetative mass (Cheplick 1989). From these 290 Cheplick ALISO

Fig. 4.—Relation of CH/CL to light intensity for six populations of Dichanthelium clandestinum in New Jersey, USA. Data were calculated from mean values for % CH and % CL reported in Bell and Quinn (1987). because biomass was significantly lower whenever soil moisture was limited (Bell and Quinn 1987). Populations showed variable patterns of plasticity in CL allocation along the moisture gradient. The two populations Fig. 5, 6.—Relation of CH and CL allocation, and CH/CL, to that showed a relatively high allocation to CL at low soil population source in Microstegium vimineum. Bars are x Ϯ SE; N moisture were from sites where soil moisture was typically ϭ 20 (except for edge where N ϭ 15).—5. Percent allocation of low, suggesting increased CL as a possible adaptation to dry tiller vegetative mass to CH and CL.—6. Ratio of CH to CL. conditions. In the light intensity experiment, populations of D. clan- ratio, despite being collected from a population with a very destinum differed significantly in allocation to CH and CL high intraspecific density (Cheplick 2005). within and across all light treatments (Bell and Quinn 1987). Because the three tiller groups differed greatly in size, Hence, there were genetically based differences in popula- allometric relationships of CH and CL allocation to vege- tion allocation patterns to CH and CL. Ratios of CH/CL tative mass (VM) were plotted (Fig. 7, 8). For shaded tillers were calculated from the allocation data (Fig. 4) in Bell and there was a significant decrease in % CH with increasing Quinn (1987). Populations varied from no effect of light on VM (log [% CH] ϭ 1.64 Ϫ 0.44 log VM; r2 ϭ 0.28, P Ͻ CH/CL (population 3 in Fig. 4), a decrease in CH/CL with 0.05), but this relationship was insignificant for edge (r2 ϭ increasing light (population 6), and an increase in CH/CL 0.13, P Ͼ 0.05) and greenhouse (r2 ϭ 0.03, P Ͼ 0.05) tillers. with increasing light (population 5). Population 5 was the In contrast, there was a highly significant decrease in % CL only example of the predicted increase in CH under high with increasing VM for greenhouse tillers (log[% CL] ϭ resource conditions (Fig. 4). 3.33 Ϫ 0.92 log VM; r2 ϭ 0.50, P Ͻ 0.01). Percent CL and For the three populations of the invasive annual Micros- VM were not correlated in edge (r2 ϭ 0.11, P Ͼ 0.05) or tegium vimineum, mean (Ϯ SE) vegetative mass for the shaded (r2 ϭ 0.14, P Ͼ 0.05) tillers. Due to these allometric greenhouse-, edge-, and shade-reared tillers was 796.2 Ϯ patterns, CH/CL was positively related to VM in the green- 56.3 mg, 281.2 Ϯ 25.8 mg, and 140.1 Ϯ 17.8 mg, respec- house tillers (log[CH/CL] ϭϪ2.59 ϩ 0.77 log VM; r2 ϭ tively. The greenhouse tillers had the lowest allocation to 0.29, P Ͻ 0.01). Thus, the largest plants in the most re- both CH (2.1 Ϯ 0.1%) and CL (5.1 Ϯ 0.4%) (Fig. 5). Tillers source-rich environment preferentially allocated more to CH from the sunny edge habitat had the highest allocation to CH relative to CL, in agreement with the idea of CH as an op- (7.7 Ϯ 0.9%) and CL (15.1 Ϯ 1.3%). Tillers from the shaded portunistic mode of reproduction. forest interior allocated a similar amount of biomass to CH (6.0 Ϯ 0.6%) and CL (6.9 Ϯ 0.7%). Interestingly, CH/CL Phenology was lowest in the resource-rich greenhouse and did not differ from that of field tillers from the edge habitat (Fig. 6). How- In the previous studies of CH and CL allocation patterns, ever, tillers from the shaded habitat had the highest CH/CL data were collected at a single point in time, typically when VOLUME 23 Chasmogamy/Cleistogamy in Grasses 291

3.3 to 9.1% in the garden and from 1.9 to 3.3% in the ﬁeld (Cheplick and Clay 1989). In the rhizomatous perennial Amphibromus scabrivalvis, a population collected from Louisiana, USA, was reared outdoors in Bloomington, Indiana, USA. Reproductive tillers were sampled at three stages (Cheplick and Clay 1989): early (before terminal CH panicles had emerged), mid (at the time of terminal panicle maturation), and late (one month later). Percent allocation to CL increased from 7.0 to 14.3% while CH allocation increased from 0 to 20.6%. Hence, CH/ CL showed a pronounced increase between early and late developmental stages (Fig. 11). Data on CH and CL allocation over time were also ex- tracted for the annual amphicarpic grass Amphicarpum purshii from Appendix A in Cheplick (1982). Fifteen plants were harvested from a ﬁeld population 2 km south of Lake- hurst in the Pinelands of New Jersey, USA, once every two weeks throughout the summer growing season. In this species, CL spikelets and seeds are matured on subterranean culms throughout the summer, while aerial spikelets and seeds on open panicles are produced only in late summer or early autumn. Between 16 and 30 Jul, CL allocation increased from 12.1 to 29.1%, but CH/CL was zero because CH panicles were not yet present (Fig. 12). Between 13 Aug and 10 Sep, CH allocation increased from 1.3 to 6.8%. Therefore, CH/CL increased substantially from 0.03 to 0.16 during this period (Fig. 12).

DISCUSSION

Chasmogamous and cleistogamous reproductive allocation in grasses with mixed breeding systems varies greatly among species and populations within species. Variation

Fig. 7, 8.—Allometry of percent CH and CL allocation (log10– among populations is not surprising given the plasticity of log10 plots) for three population sources in Microstegium vimineum CH and CL in relation to environmental factors such as light, from central New Jersey, USA. Tillers were from the shaded woods soil moisture and fertility, and competitive stress. Within nat- (᭡), sunny edge habitat (Ⅺ), or greenhouse (⅜).—7. Percent allo- ural populations, underlying all of the environmental influ- cation of tiller vegetative mass to CH.—8. Percent allocation of tiller ences on CH/CL is variation in the plasticity of genotypic vegetative mass to CL. responses. Genetically based differences in the phenotypic plasticity of CH and CL across populations can be detected by common garden experiments (Fig. 4; Bell and Quinn plants were mature and had produced seeds in both CH and 1987). Variation in CL among seed-derived sibships (Clay CL spikelets. However, the phenology of CH and CL flow- 1982; Cheplick and Quinn 1988) or cloned genotypes (Che- ering can vary greatly between species, further complicating plick 1995) reveal that the raw material for population mi- research into the evolution of mixed breeding systems. As croevolution exists in the few CL grasses examined to date. an example, contrasting temporal patterns of CH and CL In addition to environmental and genetic considerations, reproduction are presented for three cleistogamous grasses. temporal changes during development (ontogeny) and the In the caespitose perennial Danthonia spicata, apical CH life cycle (phenology) can contribute to observed variation flowering precedes CL flowering and axillary CL seeds con- in the balance of CH and CL, as depicted by the CH/CL tinue to mature at the upper nodes as plants age (Cheplick ratio. The mass of vegetative parts such as leaves, stems, and Clay 1989). In a population originally collected from and roots increases during the development of both annual Durham, North Carolina, USA, but maintained in a garden and perennial grasses as tillers increase in size and number. plot in Bloomington, Indiana, USA, CH/CL dropped from Given an allometric relationship between vegetative mass 2.8 to 1.1 over a three-week period (Fig. 9). For another and % CH or % CL, any environmental factor that slows (or population collected from the field in Monroe County, In- speeds) development and changes plant size can indirectly diana, USA, CH/CL was 7.9 in mid-Jun, but declined to 3.3 alter the CH/CL ratio (Fig. 3). The drastic reduction in size by Jul (Fig. 10). This drop in CH/CL is predominantly due associated with increasing intraspecific competition in Am- to an increase in CL allocation over time; CH allocation only phicarpum purshii (Table 1) and the relative decrease in % varied between 9.3 and 11.0% in the garden population and CH (but not % CL) explains the decrease in CH/CL as den- decreased from 15.1 to 13.3% in the field population over sity increases (Fig. 2). five weeks. At the same time, CL allocation increased from The theoretical expectation that the relative proportion of 292 Cheplick ALISO

Fig. 9–12.—Contrasting patterns in the phenology of CH and CL in three grasses over a single growing season.—9. Danthonia spicata in a garden plot in Bloomington, Indiana, USA (N ϭ 15, 14, and 13 for 7 Jun, 15 Jun, and 21 Jun, respectively).—10. Danthonia spicata from a field site in Monroe County, Indiana (N ϭ 30 per date).—11. Amphibromus scabrivalvis from Louisiana, USA, but reared outdoors in Bloomington, Indiana. ‘‘Early’’ is before terminal CH panicle emergence (N ϭ 10), ‘‘mid’’ is at the time CH panicles were mature (N ϭ 16), and ‘‘late’’ is one month later (N ϭ 10).—12. Amphicarpum purshii from a field site in the Pinelands of New Jersey, USA (N ϭ 15 per date). VOLUME 23 Chasmogamy/Cleistogamy in Grasses 293 the reproductive allocation devoted to CL would be greatest Conclusions under environmentally stressful conditions is restricted to Identifying the selective forces responsible for the evo- certain grasses exposed to specific conditions. This expec- lution of mixed breeding systems in the grasses has been tation is certainly not universal and cannot be unequivocally confounded by the tremendous phenotypic plasticity found applied to all CL grasses studied to date. In A. purshii, most among genotypes, populations, and species. It is clear that reproductive allocation was to CL at high density (Fig. 2). environmental factors, both abiotic and biotic, can alter the However, low soil nutrient availability did not significantly balance between CH and CL and that some proportion of change the CH/CL ratio in this species, despite the smaller the phenotypic changes in reproductive allocation can be at- size of plants relative to those in higher nutrient conditions tributed to changes in plant size. However, the ecological (Cheplick 1989). Likewise, in Triplasis purpurea, CH/CL and evolutionary consequences of plasticity in CH and CL was not significantly changed by decreasing soil fertility (Ta- allocation have not been thoroughly characterized. ble 2). In Dichanthelium clandestinum, the proportion of the Population genetic structure is strongly affected by the reproductive allocation composed of CL spikelets increased highly inbred nature of CL breeding systems (Sun 1999; with soil moisture stress, in agreement with theoretical ex- Green et al. 2001; Lu 2002). Although within-population pectations; however, there was no consistent effect of light molecular genetic diversity may be low, high levels of ge- on CH/CL and the ratio was greatest under low light con- netic differentiation can occur between populations (Godt ditions for some populations (Fig. 4; Bell and Quinn 1987). and Hamrick 1998; Sun 1999). It is likely that such species Finally, data for Microstegium vimineum revealed that the exist as genetically different inbred lines that occasionally lowest CH/CL ratios were obtained for the largest plants in outcross (Green et al. 2001; Lu 2002). Significant quantita- the most resource-rich environment (Fig. 7, 8)! tive genetic variation in life-history traits occurs among sib- In short, the proximate reason for plasticity in the CH/CL ships in highly inbred species (Clay 1982; Cheplick and balance across variable environments can be exposed when Quinn 1988; Charlesworth and Charlesworth 1995). Hence, the allometry of reproductive allocation is explored. Size the evolutionary potential of CL grasses may be consider- dependence of both CH and CL has been recognized in other able. herbaceous plants with mixed breeding systems (Wilken Further research is needed to test the models for the evo- 1982; Jasieniuk and Lechowicz 1987; Cheplick 1994; Berg lution and maintenance of CL, as noted over 20 years ago and Redbo-Torstensson 1998; Diaz and Macnair 1998). in the review by Campbell et al. (1983). The assumption that Clearly, future studies of putative adaptive responses of CH CL mostly provides reproductive assurance should be re- and CL to environmental conditions should not ignore the evaluated in light of information on allometric relationships role of size in mediating resource partitioning to both repro- and phenological patterns. Characterization of the molecular ductive modes. and quantitative genetic variation within and among popu- The patterns detected in the reproductive allocation to CH lations, and the plasticity of life-history traits, will be nec- and CL for a particular species will also be closely linked essary to further refine evolutionary models and understand to flowering phenology, which may be relatively fixed and the ecological success of species with mixed breeding sys- not necessarily related to size. When CH precedes CL, a tems. seasonal decline in CH/CL is expected (Fig. 9, 10), while the reverse is expected when CH follows CL (Fig. 11, 12). ACKNOWLEDGMENTS The phenological timing may be especially important in an- Thanks are extended to J. A. Quinn for the invitation to nual herbs of disturbed environments where the risks of early participate in the symposium and to two anonymous review- mortality can be high (Bazzaz 1996). For example, because ers for comments on the manuscript. Amphicarpum purshii allocates resources to CL much earlier in the growing season than it does to CH (Fig. 12), early LITERATURE CITED death might mean that an individual has reproduced, but only via seeds matured in CL spikelets. The consistency in BAZZAZ, F. A. 1996. Plants in changing environments. Cambridge CH/CL phenology of species such as A. purshii provides a University Press, Cambridge, UK. 320 p. solid argument for an evolutionarily fixed strategy of repro- BELL, T. J., AND J. A. QUINN. 1987. Effects of soil moisture and light duction. To date, it is not known to what extent mortality intensity on the chasmogamous and cleistogamous components of reproductive effort of Dichanthelium clandestinum populations. risks have contributed to the evolution of this particular phe- Canad. J. Bot. 65: 2243–2249. nological pattern in A. purshii, or any other annual CL spe- BERG, H., AND P. REDBO-TORSTENSSON. 1998. Cleistogamy as a bet- cies. For perennials in which seed recruitment is compara- hedging strategy in Oxalis acetosella, a perennial herb. J. Ecol. tively rare, early production of seeds by CL to provide re- 86: 491–500. productive assurance may not be a tenable evolutionary hy- CAMPBELL, C. S. 1982. Cleistogamy in Andropogon L. (Gramineae). pothesis (e.g., Berg and Redbo-Torstensson 1998). Amer. J. Bot. 69: 1625–1635. Clearly, when trying to compare species or different ex- , J. A. QUINN, G. P. CHEPLICK, AND T. J. BELL. 1983. Cleis- perimental treatments within a species, investigators must be togamy in grasses. Annual Rev. Ecol. Syst. 14: 411–441. CHARLESWORTH, D., AND B. CHARLESWORTH. 1995. Quantitative ge- careful to only compare data recorded at equivalent pheno- netics in plants: the effect of the breeding system on genetic var- logical stages. Researchers should also be well aware of the iability. Evolution 49: 911–920. relative timing of CH and CL reproduction for the species CHEPLICK, G. P. 1982. The role of differential allocation to aerial under study and how phenology might impact experimental and subterranean propagules in the population dynamics and sur- results. vival strategies of Amphicarpum purshii Kunth. M.S. thesis, De- 294 Cheplick ALISO

partment of Biological Sciences, Rutgers University, New Bruns- movement and tissue specialization. Ann. Bot. (Oxford) 77: 573– wick, New Jersey, USA. 81 p. 582. . 1989. Nutrient availability, dimorphic seed production, and HORTON, J. L., AND H. S. NEUFELD. 1998. Photosynthetic responses

reproductive allocation in the annual grass Amphicarpum purshii. of Microstegium vimineum (Trin.) A. Camus, an exotic C4, shade- Canad. J. Bot. 67: 2514–2521. tolerant grass, to variable light environments. Oecologia 114: 11– . 1994. Life history evolution in amphicarpic plants. Pl. Spec. 19. Biol. 9: 119–131. HUNT, D., AND R. ZAREMBA. 1992. The northeastern spread of Mi- . 1995. Plasticity of seed number, mass, and allocation in crostegium vimineum (Poaceae) into New York and adjacent clones of the perennial grass Amphibromus scabrivalvis. Int. J. Pl. states. Rhodora 94: 167–170. Sci. 156: 522–529. JASIENIUK, M., AND M. J. LECHOWICZ. 1987. Spatial and temporal . 1996. Cleistogamy and seed heteromorphism in Triplasis variation in chasmogamy and cleistogamy in Oxalis montana (Ox- purpurea (Poaceae). Bull. Torrey Bot. Club 123: 25–33. alidaceae). Amer. J. Bot. 74: 1672–1680. . 2005. Biomass partitioning and reproductive allocation in KURAUCHI, N., T. MAKINO, AND S. HIROSE. 1993. Inheritance of cleis- the invasive cleistogamous grass Microstegium vimineum: influ- togamy-chasmogamy in barley. Barley Genetics Newsletter 23: ence of the light environment. J. Torrey Bot. Soc. 132: 214–244. 19. , AND K. CLAY. 1989. Convergent evolution of cleistogamy LLOYD, D. G. 1984. Variation strategies of plants in heterogeneous and seed heteromorphism in two perennial grasses. Evol. Trends environments. Biol. J. Linn. Soc. 21: 357–385. Pl. 3: 127–136. . 1987. Allocations to pollen, seeds, and pollination mecha- nisms in self-fertilizing plants. Funct. Ecol. 1: 83–89. , AND J. A. QUINN. 1983. The shift in aerial/subterranean fruit ratio in Amphicarpum purshii: causes and significance. Oecologia LORD, E. M. 1981. Cleistogamy: a tool for the study of floral mor- phogenesis, function and evolution. Bot. Rev. (Lancaster) 47: 57: 374–379. 421–449. , AND . 1988. Quantitative variation of life-history LU, Y. 2002. Why is cleistogamy a selected reproductive strategy in traits in amphicarpic peanutgrass (Amphicarpum purshii) and its Impatiens capensis (Balsaminaceae)? Biol. J. Linn. Soc. 75: 543– evolutionary significance. Amer. J. Bot. 75: 123–131. 553. CHHABRA, A. K., AND S. K. SETHI. 1991. Inheritance of cleistogamic MASUDA, M., T. YAHARA, AND M. MAKI. 2001. An ESS model for flowering in durum wheat (Triticum durum). Euphytica 55: 147– the mixed production of cleistogamous and chasmogamous flow- 150. ers in a facultative cleistogamous plant. Evolutionary Ecology Re- CLAY, K. 1982. Environmental and genetic determinants of cleistog- search 3: 429–439. amy in a natural population of the grass Danthonia spicata. Evo- MAZER, S. J., AND G. LEBUHN. 1999. Genetic variation in life-history lution 36: 734–741. traits: heritability estimates within and genetic differentiation . 1983a. Variation in the degree of cleistogamy within and among populations, pp. 85–171. In T. O. Vuorisalo and P. K. Mu- among species of the grass Danthonia. Amer. J. Bot. 70: 835– tikainen [eds.], Life history evolution in plants. Kluwer Academic 843. Publishers, Dordrecht, The Netherlands. . 1983b. The differential establishment of seedlings from MERWINE, N. C., L. M. GOURLEY, AND K. H. BLACKWELL. 1981. chasmogamous and cleistogamous flowers in natural populations Inheritance of papery glume and cleistogamy in sorghum. Crop of the grass Danthonia spicata (L.) Beauv. Oecologia 57: 183– Sci. (Madison) 21: 953–956. 188. NIKLAS, K. J. 1994. Plant allometry: the scaling of form and process. CULLEY, T. M. 2002. Reproductive biology and delayed selfing in The University of Chicago Press, Chicago, Illinois, USA. 395 p. Viola pubescens (Violaceae), an understory herb with chasmoga- PLITMANN, U. 1995. Distribution of dimorphic flowers as related to mous and cleistogamous flowers. Int. J. Pl. Sci. 163: 113–122. other elements of the reproductive strategy. Pl. Spec. Biol. 10: 53– DIAZ, A., AND M. R. MACNAIR. 1998. The effect of plant size on the 60. expression of cleistogamy in Mimulus nasutus. Funct. Ecol. 12: REDBO-TORSTENSSON, P., AND H. BERG. 1995. Seasonal cleistogamy: 92–98. a conditional strategy to provide reproductive assurance. Acta Bot. Neerl. 44: 247–256. DOBRENZ, A. K., AND A. A. BEETLE. 1966. Cleistogenes in Dan- thonia. J. Range Managem. 19: 292–296. SCHEMSKE, D. W. 1978. Evolution of reproductive characteristics in Impatiens (Balsaminaceae): the significance of cleistogamy and DYKSTERHUIS, E. J. 1945. Axillary cleistogenes in Stipa leucotricha chasmogamy. Ecology 59: 596–613. and their role in nature. Ecology 26: 195–199. SCHOEN, D. J. 1984. Cleistogamy in Microlaena polynoda (Grami- GIBSON, D. J., G. SPYREAS, AND J. BENEDICT. 2002. Life history of neae): an examination of some model predictions. Amer. J. Bot. Microstegium vimineum (Poaceae), an invasive exotic in southern 71: 711–719. Illinois. J. Torrey Bot. Soc. 129: 207–219. , AND D. G. LLOYD. 1984. The selection of cleistogamy and GODT, M. J. W., AND J. L. HAMRICK. 1998. Allozyme diversity in heteromorphic diaspores. Biol. J. Linn. Soc. 23: 303–322. the grasses, pp. 11–29. In G. P. Cheplick [ed.], Population biology SUN, M. 1999. Cleistogamy in Scutellaria indica (Labiatae): effec- of grasses. Cambridge University Press, Cambridge, UK. tive mating system and population genetic structure. Molec. Ecol. GREEN, J. M., J. H. A. BARKER, E. J. P. MARSHALL, R. J. FROUD- 8: 1285–1295. WILLIAMS, N. C. B. PETERS, G. M. ARNOLD, K. DAWSON, AND A. UENO, K., AND H. ITOH. 1997. Cleistogamy in wheat: genetic control KARP. 2001. Microsatellite analysis of the inbreeding grass weed and the effect of environmental conditions. Cereal Res. Commu- Barren Brome (Anisantha sterilis) reveals genetic diversity at the nications 25: 185–189. within- and between-farm scales. Molec. Ecol. 10: 1035–1045. WILKEN, D. H. 1982. The balance between chasmogamy and cleis- HESLOP-HARRISON, Y., AND J. S. HESLOP-HARRISON. 1996. Lodicule togamy in Collomia grandiflora (Polemoniaceae). Amer. J. Bot. function and filament extension in the grasses: potassium ion 69: 1326–1333.